Advanced search
Start date
Betweenand

Structured stochastic processes and functional data analysis for the assessment of motor learning in normal and pathological subjects

Grant number: 16/22053-7
Support type:Scholarships in Brazil - Post-Doctorate
Effective date (Start): February 01, 2017
Effective date (End): November 30, 2020
Field of knowledge:Physical Sciences and Mathematics - Probability and Statistics
Principal Investigator:Jefferson Antonio Galves
Grantee:Noslen Hernández González
Home Institution: Instituto de Matemática e Estatística (IME). Universidade de São Paulo (USP). São Paulo , SP, Brazil
Associated research grant:13/07699-0 - Research, Innovation and Dissemination Center for Neuromathematics - NeuroMat, AP.CEPID

Abstract

Neuromat is a research center aimed at developing a Theory of Brain. Recent results involves the definition of new kind of stochastic processes, namely stochastic processes driven by context tree models, that allow for the design, modeling and analysis of neurophysiological experiments with structured stimuli. Stochastic processes driven by context tree models have shown to be a useful tool to address the hypothesis that the brain retrieves statistical regularities from stimuli. First results on this direction are presented in a publication of Duarte et al. (ArXiv: 1602.00579) where it was shown that EEG data encodes structure stimuli the subjects were submitted to. An open question is whether such finding could also be corroborated in behavioral responses, specifically in the execution of movements. The main purpose of the present project is to develop and apply methods addressing this question on the basis of the general mathematical setting just mentioned. In particular, these methods are aimed to contribute to the study of brain functioning in the execution of movements in rehabilitation of patients, and the Goalkeeper Game. For this, stochastic process driven by structured Markov Chains needs to be considered in a variety of spaces, including categorical responses as well as responses that are functional data (such as curves of motions). This also poses a number of challenges to be dealt with in the representation, computer processing and pattern recognition of complex recorded signals that corresponds to gestures, postures and movements. The project is structured in two main research lines: 1) Stochastic processes driven by structured Markov Chains for the assessment of motor learning and 2) Functional data analysis of motion curves. In the first research line, several challenging sub-problems will be studied like the design of suitable experiments to evaluate evidence about learning statistical regularities; the mapping from raw data to relevant response features and; Inference. In the second research line modern functional data analysis tools will be applied in a number of practical situations concerning kinematics of behavior in which several relevant aspects of the response of subjects can be represented by functional data. (AU)